© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    3. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    5. Gemma-2-9B-IT
    7. Residual Stream - 131k
    9. 31-GEMMASCOPE-RES-131K
    11. 65272
    Prev
    Next
    INDEX
    Explanations

    instances of verbs in various tenses, especially forms of "is," "are," "was," and "were."

    oai_token-act-pair · gpt-4o-miniTriggered by @bot

    past tense verbs

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-9B-IT @ 31-gemmascope-res-131k
    Configuration
    google/gemma-scope-9b-it-res/layer_31/width_131k/average_l0_109
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    131,072
    Data Type
    float32
    Hook Name
    blocks.31.hook_resid_post
    Hook Layer
    31
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     réalis
    -0.54
     representa
    -0.47
     océ
    -0.46
     inspira
    -0.44
     desapar
    -0.40
     معت
    -0.40
     informar
    -0.40
     plufieurs
    -0.40
     Italij
    -0.40
    Compos
    -0.40
    POSITIVE LOGITS
     taken
    1.18
     brought
    1.10
     kept
    0.96
     transferred
    0.95
     pushed
    0.94
     handled
    0.94
     placed
    0.93
     examined
    0.93
     thrown
    0.91
     sent
    0.91
    Activations Density 0.838%

    No Known Activations